Why version control?

Two examples in which proper version control can be a life/time saver

Version control with Git

• Complete and long-term history of every file in your project

• Open source and free to use version control software

• Quasi standard for software development

• A whole universe of other software and services around it

The basic idea of Git

• For projects with mainly text files (e.g. code, markdown files, …)

• Basic idea: Take snapshots (commits) of your project over time

• A project version controlled with Git is a Git repository (repo)

Version control with Git

Git is a distributed version control system

How to use Git

After you installed it there are different ways to interact with the software.

How to use Git - Terminal

Using Git from the terminal

How to use Git - Integrated GUIs

A Git GUI is integrated in most (all?) IDEs, e.g. R Studio, VS Code

How to use Git - Standalone GUIs

Standalone Git GUI software, e.g. Github Desktop

How to use Git

Which one to choose?

• Depends on experience and taste
• You can mix methods because they are all interfaces to the same Git

Step 0: An empty project

Step 1: Initialize a git repository

• Adds a (hidden) .git folder to your project
  • You don’t need to interact with this folder directly

Step 2: Modify files and stage changes

Git detects any changes in the working directory

Step 2: Modify files and stage changes

Stage file to be part of the next commit (snapshot)

• In the terminal use git add
• In GUIs just a check box

Step 3: Commit changes

• Commits are the snapshots of your project state
• Commit a bundle of changes from staging area to local repo
• Collect meaningful chunks of work in the staging area, then commit

Step 3: Commit changes

• After a commit, the staging area is clear again
• Changes are now part of the project’s Git history

How to write good commit messages?

How to write good commit messages?

See here for more details but some general rules:

1. Limit summary line to 50 characters
2. Capitalize summary line
3. Do not end summary line with period
4. Use imperative mood in the subject line
5. Use the Description to explain what and why, not how

How to write good commit messages?

Limit model temperature range

I modified the temperature range the model can 
operate on to an upper limit of 40°C. This 
fixes the following problems:

- no more unrealistic results because 
temperature cannot exceed the meaningful range
- the program does not terminate with error 
code 123 anymore

limited model temperature range.

Temperatures above 40°C are unrealistic.

The commit history

The commit history

The commit history

Step 4: Create and connect a remote repo

• Use remote repos (on a server) to synchronize, share and collaborate

• Remote repos can be private (you + collaborators) or public (visible to anyone)

Step 5: Share changes with the remote repo

• Push your local changes to the remote with git push

A word on remote repositories

• There are commercial and self-hosted options for your remote repositories
  • Commercial: Github, Gitlab, Bitbucket, …
  • Self-hosted: Gitlab (maybe at your institution?)
• For the commercial options please be aware of your institutional guidelines
  • Servers are likely outside EU
  • Privacy rules might apply

Summary of the basic steps

Checkout a previous commit

• Bring your work space back in time temporarily with git checkout

Revert changes

• Use git revert to revert specific commits
• This does not delete the commit, it creates a new commit that undoes a previous commit
  • It’s a safe way to undo commited changes

Ignore files with .gitignore

• Use a .gitignore file to list files/folders that you don’t want to track with git

Ignore files with .gitignore

• Create a file with the name .gitignore in working directory

• Add all files and directories you want to ignore to the .gitignore file

*.html    # ignore all .html files
*.pdf     # ignore all .pdf files

debug.log # ignore the file debug.log

build/    # ignore all files in subdirectory build

Publish your repositories

Github/Gitlab are a good way to publish and share your work.

Publish your repositories

You can increase the quality/complexity of your repo by

• Adding a README.md file
• Connecting the repo with Zenodo to get a DOI
• Create a Github pages website alongside your repo
• Encourage people to write issues if they find problems
• …